MiniMax's long-term outlook is constructive, with Open Platform, agent usage, and multimodality as the key drivers

Overview

This report summarizes Morgan Stanley's feedback on MiniMax after China Summit 2026. The core view is that MiniMax has rapid model iteration, strong overseas demand, agent usage growth, multimodal expansion, and room for cost optimization, making the long-term outlook constructive. In the near term, the key focus is expanding global go-to-market capabilities to capture inbound overseas demand, while continuing to reduce inference costs and enhance multimodal capabilities.

Core views

Open Platform is the core growth engine, contributing about 40% of total ARR. Its customer base includes global enterprises, developers, and individual users, and most major Chinese internet companies are customers. Overseas Open Platform revenue has exceeded domestic revenue since 4Q25, indicating strong international demand. Management views agent usage as a structural driver because it executes continuous multi-step tasks and consumes token volumes far higher than those of a single-turn interactive chatbot. Multimodal and generative media also contribute about 40% of ARR, and as the model expands from professional users and prosumers to mass users, it could become the next catalyst for scale.

Analysis framework

The report evaluates management feedback, business ARR mix, changes in overseas demand, the agent adoption path, multimodal product usage, model iteration speed, and room for inference-cost optimization, and uses a DCF valuation framework to derive the target price.

Methodology notes

• Valuation methodsDCF

  Discounted cash flow valuation

  Morgan Stanley uses the DCF method, assuming a WACC of 15% and a perpetual growth rate of 3%; the target price implies about 56x P/S for 2027e.

• research_frameworkMorgan Stanley ModelWare

  Internal forecasting framework

  Unless otherwise stated, the report's metrics are based on the Morgan Stanley ModelWare framework, research estimates, and Refinitiv consensus data.

• rating_systemMorgan Stanley Stock Rating

  Relative rating system

  Overweight means that the stock's total return over the next 12-18 months is expected to outperform the average total return of the analyst-covered industry on a risk-adjusted basis.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• MiniMax (0100.HK)

  Core covered name

  Strengths

  Open Platform and generative media/multimodal each contribute about 40% of ARR; overseas demand is strong; model iteration is fast; the founder CEO and CTO are deeply involved in frontier model R&D; cash burn is low and R&D efficiency is relatively high.

  Weaknesses

  The global sales team remains relatively lean; multimodal adoption is currently more concentrated among professional users and prosumers; profitability metrics and EPS information are limited.

  Comparison

  Compared with hyperscale cloud vendors, management emphasizes that MiniMax has higher R&D efficiency and lower cash burn; compared with traditional chatbots, agent usage can drive higher token consumption.

  Risks

  Insufficient overseas go-to-market expansion, slower-than-expected declines in inference costs, weaker-than-expected mass-market adoption of multimodal products, and high valuation multiples.

Key data

• Open Platform ARR shareabout 40%Open Platform serves global enterprises, developers, and individual users, and is the core growth engine.
• Open Platform overseas revenueexceeded domestic revenue since 4Q25Management believes the strong inbound demand reflects product pull rather than sales push.
• Generative media and multimodal ARR shareabout 40%Current adoption is mainly concentrated among professional users and prosumers, and the next-generation model may expand into the mass market.
• Target priceHK$990.00The target price implies about 56x P/S for 2027e.
• Closing priceHK$929.50As of 2026-03-31.
• Implied upside7%Calculated based on the target price and closing price.
• WACC15%DCF valuation assumption.
• Perpetual growth rate3%DCF valuation assumption.
• Revenue forecastUS$79mn / US$240mn / US$712mn / US$1,531mnCorresponding to 12/25, 12/26e, 12/27e, and 12/28e, respectively.

Impact & implications

If overseas demand continues to scale, agent usage penetration accelerates, and inference costs decline, MiniMax's ARR and revenue growth could show strong upside elasticity. If multimodal models improve ease of use and understanding capabilities, they may also expand the user base from professionals to the mass market. However, the company still needs to prove that its global sales and delivery capabilities can match inbound demand, while continuing to optimize inference costs.

Risks

• An insufficient global go-to-market capability may prevent the company from capturing overseas inbound demand.
• Inference costs, especially multimodal inference costs, may not improve as expected, which could pressure scaling and margins.
• The current usage barrier for multimodal products is relatively high; if the next-generation model fails to meaningfully improve ease of use, user expansion may be constrained.
• The growth path for agent usage is still at an early penetration stage, and the pace of organizational and industry-level deployment remains uncertain.
• The target price implies about 56x P/S for 2027e, making the valuation sensitive to long-term growth and execution delivery.
• Morgan Stanley discloses an investment banking relationship and potential conflicts of interest with MiniMax; investors should treat this report as one decision factor rather than the sole basis.

What to watch

• Whether overseas Open Platform revenue continues to exceed domestic revenue and sustain growth.
• Whether the expansion of global sales and go-to-market teams matches inbound demand.
• Whether agent usage expands from individual users into internal departments such as HR, finance, and legal, and then further into industry-level deployments.
• Whether multimodal models achieve better context understanding and generation capabilities and lower the usage barrier for mass-market users.
• Progress in optimizing inference costs, especially multimodal inference costs.
• Whether subsequent revenue, ARR, cash burn, and profitability path support the high valuation.